K+ and Cl− channels/transporters independently fine-tune photosynthesis in plants

In variable light environments, plants adjust light use in photosynthetic electron transport and photoprotective dissipation in the thylakoid membrane. In this respect, roles of the K+/H+ antiporter KEA3, the Cl− channel/transporter CLCe and the voltage-dependent Cl− channel VCCN1 have been unraveled in Arabidopsis thaliana. Here we report that they independently adjust photosynthesis on the basis of analyses using single and higher order loss-of-function mutants. In short experiments of photosynthetic response on transition from dark to low light, we reveal a sequential functioning of VCCN1 and CLCe in the activation of photoprotection and of KEA3 in its downregulation to a low steady state while adjusting the electron transport. On transition from low to high light, VCCN1 accelerates the activation of photoprotection, whereas KEA3 slows it down on transition from high to low light. Based on parallel electrochromic band shift measurements, the mechanism behind is that VCCN1 builds up a pH gradient across the thylakoid membrane, whereas KEA3 dissipates this gradient, which affects photoprotection. CLCe regulates photosynthesis by a pH-independent mechanism likely involving Cl− homeostasis. Nevertheless, all genotypes grow well in alternating high and low light. Taken together, the three studied ion channels/transporters function independently in adjusting photosynthesis to the light environment.

Plants harvest sunlight and convert the captured energy into biochemical products. For the energy conversion to take place, several processes occur in a timely manner in what is generically called photosynthetic light-dependent reactions. Sunlight is not only the energy source for photosynthesis but also an important limiting factor due to its fluctuations in intensity 1 . Rapid and abrupt changes in light conditions affect electron transport through photosystem II (PSII), cytochrome b 6 f and photosystem I (PSI) complexes in the thylakoid membrane 2 . To overcome the constraints imposed by sunlight, energy capture and dissipation of excess light energy are tightly regulated during photosynthesis. In order to avoid damage to photosystems, excess energy is quenched by dissipation as heat in a mostly pH-dependent process known as qE-type of non-photochemical quenching (NPQ) 3 . NPQ undergoes activation and downregulation in response to changes in light intensity, and is controlled by the pH in the thylakoid lumen, the pH sensor protein PsbS and the carotenoid zeaxanthin 4 . The rates of NPQ activation and in particular of its downregulation are rather slow (order of minutes) as compared to natural light fluctuations (order of seconds), and are likely limited by the slow kinetics of pH changes in the thylakoid lumen 5 . Several Arabidopsis thaliana npq mutants have been discovered, but they grew similarly to wild-type plants under constant laboratory light conditions 6,7 . Nevertheless, experiments in field conditions with the same mutants indicated that NPQ confers increased plant tolerance to fluctuating light 1 . Moreover, tobacco mutants engineered to accelerate NPQ response to fluctuating light showed improved photosynthesis and biomass production 8 .
There is increasing evidence that NPQ is also regulated by ion channels/transporters able to modulate the kinetics of pH changes in the lumen 9,10 . In the thylakoid membrane of Arabidopsis, the two-pore K + channel TPK3 is essential for activation of NPQ 11 , whereas the K + /H + antiporter KEA3 accelerates its downregulation and

Results
Growth and photosynthesis phenotype under standard light conditions. To better understand the role of CLCe, KEA3 and VCCN1 and their possible interdependency in the regulation of PMF and photosynthesis, we compared the phenotypes of clce-2, kea3-1 and vccn1-1 single loss-of-function mutant lines 14,17,18 , hereafter named clce, kea3 and vccn1, with those of newly generated clcekea3, clcevccn1, kea3vccn1 double and clcekea3vccn1 triple mutant lines.
All mutant lines grew similarly to wild-type (wt) plants under standard light conditions (120 µmol photons m −2 s −1 ) and displayed similar shoot fresh weight at 5 weeks (Fig. 1a,b) and 8 weeks ( Supplementary Fig. 1a,b). The chlorophyll (Chl) content and Chl a/b ratio were also similar to wt at both ages (Table 1 and Supplementary  Table 1).
To determine the maximal photosynthetic efficiency in terms of electron transport, we recorded fast Chl fluorescence (OJIP) kinetics on intact leaves of dark-adapted plants. At 5 weeks, the curve in wt had a polyphasic shape, while lower fluorescence levels at the J and I steps, previously reported in clce 14,15 , were also reproduced in this study and retained in clce double and triple mutant lines (Fig. 2a). The curves were double normalized to J and P and the curve difference between wt and clce lines was plotted. The resulting peak corresponds to the I step ( Fig. 2a inset), associated with electron transport at PSI 22 . Double normalization to O and J reflecting PSII antenna size and activity resulted in much smaller differences between wt and the clce lines as compared to those in the I peak ( Supplementary Fig. 2). In addition, the maximal quantum yield of PSII (F v /F m ), the photosynthetic performance index (PI ABS , an overall indicator of efficiency of photosynthesis), and the time to reach maximal fluorescence intensity (t Fm ) were similar to wt in all mutants with the exception of clce lines (Table 1). F v /F m and PI ABS in clce were slightly but significantly reduced, whereas t Fm was over 2-fold higher, indicating a slower rate to close all PSII centers. This phenotype was preserved in clcekea3 and clcevccn1 double and clcekea3vccn1 triple lines as well. We also recorded oxidation-reduction kinetics of P700 (PSI primary donor) and found a more pronounced reduction of P700 + in dark-incubated clce single and clcekea3vccn1 triple mutant lines than in wt ( Supplementary Fig. 3), in line with previous findings 14 . This suggests an accelerated electron transfer between PSII and PSI and could explain the lower fluorescence levels at the J and I steps (Fig. 2a) and higher values for t Fm in clce lines ( Table 1). The observed differences are typical for the dark state since clce plants adapted to standard (growth) light were indistinguishable from wt in P700 + reduction kinetics ( Supplementary Fig. 3) as well as in the light response curves of the PSI quantum yield (Y(I), Supplementary Fig. 4).
We also recorded Chl fluorescence in plants under steady-state standard light conditions and again found that the clce lines differed from the other genotypes after 5 weeks of growth. The PSII quantum yield (Y(II)) as measure of the effective electron transport of PSII was slightly but significantly reduced under these conditions ( Table 1).
The photosynthetic phenotype of clce was observed in plants grown for 5 weeks in standard light and watered with deionized water (Table 1). Remarkably, this phenotype disappeared in older clce plants (8 weeks) as based on the non-significant differences from the wt for the Chl fluorescence parameters (Supplementary Table 1), in accordance with previous observations 10 . We further found that when tap water was used (1-5 mM Cl − ), 6-week-old clce lines showed Chl fluorescence and t Fm values indistinguishable from wt ( Supplementary Fig. 5). This may be due to increased accumulation of Cl − in leaves with age and watering conditions, which compensate for the disturbed Cl − distribution in clce lines. In support of this hypothesis, detached leaves of 5-week-old clce lines that were pre-incubated in deionized water supplemented with Cl − (150 mM) displayed normal Chl fluorescence and t Fm similar to wt (Fig. 2b,c), in accordance with previous findings 14 .
To further strengthen our hypothesis about a disturbed Cl − homeostasis in clce, in a separate set of experiments we have grown plants with different Cl − supply in the watering solution until the age of 6 and 8 weeks. The plants supplied with 0.5 mM Cl − preserved the fluorescence phenotype observed with deionized water at 6 but not at 8 weeks, whereas plants supplied with 5 mM Cl − resembled wt at both ages (Supplementary Figs. 6, 7). Since we obtained consistent fluorescence phenotypes with reproducible data when the clce plants were watered with deionized water at the age of 5-6 weeks, we used this watering procedure for all genotypes throughout the experiments in this study. Taken together, our observations from experiments on the dark-and light-adapted www.nature.com/scientificreports www.nature.com/scientificreports/ plants indicate that all lines grow well and perform wt-like photosynthesis under standard light conditions with the exception of clce. The clce lines experience an altered Cl − homeostasis which impacts photosynthetic electron transport, without, however, affecting the capacity to produce biomass.

Regulation of photosynthesis on transition from dark to light. Slow kinetics of Chl fluorescence
induction were recorded to determine the role of CLCe, KEA3 and VCCN1 in the regulation of photosynthesis. Upon transition from dark to low light, NPQ transiently decreased in vccn1 and clce, whereas in kea3 it increased 2-fold and relaxed more slowly before reaching a steady state similar to wt (Fig. 3). Double and triple mutant lines displayed intermediate NPQ patterns, indicating additive effects of the mutations. To visualize the contribution of each individual channel/transporter to the kinetics of NPQ, we calculated the difference of wt minus mutant 23 . VCCN1 accelerated NPQ activation with a maximum at 0.5 min followed by CLCe with a maximum at 1 min of illumination. KEA3 relaxed most of the NPQ during 3 min of illumination. During the overlapping periods, NPQ became the sum of individual contributions, as evidenced by the intermediate NPQ difference in the double and triple mutant lines (Fig. 3 inset, Supplementary Fig. 8).
Photosynthetic electron transport measured as PSII and PSI quantum yield (Y(II) and Y(I), respectively) in the same experiment, was reduced in kea3 during the 3 min corresponding to NPQ induction and relaxation, and was similar to wt during the steady state (Fig. 3). The clce displayed a reduced Y(I) throughout the illumination and a lower Y(II) than wt when reaching the steady state of NPQ. The corresponding parameters in vccn1 were similar to wt, whereas the double and triple mutant lines displayed intermediate patterns. Based on the determined Y(II) and Y(I) differences ( Fig. 3 insets), KEA3 and CLCe stimulated electron transport adjusting it to the NPQ on transition from dark to low light.
We have also studied the dynamics of NPQ and photosynthesis on transition from dark to high light. Under these conditions, NPQ was induced more slowly and was indistinguishable among vccn1 lines, whereas Y(II) and Y(I) were not largely affected (Fig. 4). NPQ and electron transport in kea3 and clce single and double clcekea3 www.nature.com/scientificreports www.nature.com/scientificreports/ lines were similar to wt. VCCN1 accelerated the activation of NPQ with a maximum at 0.5 min until 3 min of high light when a steady state was reached, and this effect was not altered by either CLCe or KEA3 ( Fig. 4 inset, Supplementary Fig. 8). Taken together, these data suggest a sequential and additive action of VCCN1, CLCe and KEA3 in modulating NPQ and electron transport on transition from dark to low light, and a function solely for VCCN1 in modulating these processes on transition from dark to high light.

Regulation of photosynthesis on transition from low to high light.
To further investigate the role of CLCe, KEA3 and VCCN1 in photosynthetic regulation, we undertook experiments under fluctuating light (transitions from low intensity to high intensity and back to low intensity). On transition from low to high light, NPQ was induced slower in vccn1 than in wt and reached a lower steady state (Fig. 5a). NPQ in clce and kea3 largely resembled wt, even though it was induced slightly faster in kea3. Double clcekea3 and triple clcekea3vccn1 lines displayed intermediate patterns, suggesting additive effects of the mutations. Based on the determined NPQ difference, VCCN1 accelerated NPQ activation with a peak at 0.5 min until 3 min of high light, whereas CLCe and KEA3 only weakly affected the NPQ (Fig. 5a inset, Supplementary Fig. 8). The electron transport through photosystems (Y(II) and Y(I)) was similar to wt in all mutant lines except kea3 where it was lower (Fig. 5a). Y(I) and Y(II) difference revealed that VCCN1 slightly elevated the electron transport within the first 0.5 min corresponding to the peak of NPQ activation, whereas KEA3 and CLCe slightly further increased it during the remaining illumination time (Fig. 5a insets).
The calculated halftime of NPQ induction was 1.5-2-fold higher in vccn1 lines and similar to wt in the other lines ( Fig. 5b left panel). These data are consistent with a major role of VCCN1 in NPQ dynamics on transition to high light.
In parallel experiments we determined PMF size and partitioning from electrochromic shift (ECS) recordings at the end of the transition from low to high light (Supplementary Fig. 9a left panel). The ∆pH significantly decreased in vccn1 lines, whereas in the other genotypes ∆pH was similar to wt ( Fig. 5c left panel). PMF size was elevated only in vccn1 lines (Supplementary Fig. 9a right panel). Taken together, these data indicate that on transition from low to high light, VCCN1 functions in building up a significant pH gradient to rapidly activate NPQ without largely affecting the electron transport through photosystems. CLCe and KEA3 do not play a major role in photosynthetic regulation under these conditions and only fine-tune the NPQ produced by VCCN1.

Regulation of photosynthesis on transition from high to low light. Following transition from high
to low light, photosynthesis undergoes a gradual adjustment, where KEA3 was proposed to play an important role 12 . In our fluctuating light conditions, NPQ relaxed more slowly in kea3 and kea3vccn1 lines and this effect was slightly enhanced in clcekea3 double and clcekea3vccn1 triple lines, while resembled wt in the other lines (Fig. 5a). NPQ differences revealed KEA3 as a major contributor to the relaxation of NPQ in the first min of transition to low light (Fig. 5a inset, Supplementary Fig. 8).
Y(II) and Y(I) under the same conditions were lower in kea3 and clce and further decreased in the clcekea3 double and clcekea3vccn1 triple lines (Fig. 5a). Based on the Y(II) and Y(I) differences, KEA3 stimulated the electron transport of both photosystems in the first min of transition to low light, whereas CLCe elevated photosystem activity throughout the illumination (Fig. 5a insets).
The halftime of NPQ relaxation in kea3 lines was 3-fold higher than in all other genotypes including wt ( Fig. 5b right panel), consistent with a role of KEA3 on transition from high to low light. In parallel experiments with ECS recordings at the end of the transition (Supplementary Fig. 9b left panel), only the kea3 lines displayed a significantly higher ∆pH (Fig. 5c right panel) while preserving the wt PMF size ( Supplementary Fig. 9b right  panel). Taken together, these data indicate that during the high to low light transition, KEA3 functions in the dissipation of pH gradient leading to NPQ relaxation, while enhancing PSII and PSI efficiency. CLCe also affects PSI and PSII electron transport, and this happens likely by a ∆pH-independent mechanism.  Table 1. Chlorophyll content and fluorescence parameters of 5-week-old plants watered with deionized water. Plants were grown as described in Fig. 1. Leaf Chl content and Chl a/b ratio were determined spectrophotometrically from leaf discs of 16-h dark-adapted wild-type plants (wt) and mutants following extraction in ethanol. The maximal quantum yield of PSII photochemistry (F v /F m ), the photosynthetic performance index (PI ABS ) and the time to reach F m (t Fm ) were estimated from fast kinetics of Chl fluorescence in 30 min dark-adapted plants. The PSII quantum yield (Y(II)) was determined under steady-state growth light conditions (4 h of illumination at 120 μmol photons m −2 s −1 ). Data are means ± SEM from 3 experiments (n = 10 plants per experiment). Asterisks indicate statistically significant difference in the studied parameters between wt and mutants (ANOVA, P < 0.05).
www.nature.com/scientificreports www.nature.com/scientificreports/  Fig. 10) and no significant differences were observed in biomass, Chl content and the F v /F m parameter (Supplementary Table 2). These data indicate that the fine adjustments in NPQ and electron transport do not impact plant growth in our design of fluctuating light conditions. Experiments in field-like conditions are required since they may reveal an altered tolerance of the mutants to natural fluctuating light, as it was previously observed for single vccn1 lines 16 . Thylakoid ultrastructure. The thylakoid membrane undergoes dynamic changes in its overall shape depending on the light environment, and this may control important processes in photosynthesis 19 . Recent reports highlighted CLCe and VCCN1 as important players for the thylakoid ultrastructure in the dark-and light-adapted state, respectively 14,17 . Here we studied chloroplast and thylakoid ultrastructure of clce, kea3 and vccn1 single and clcekea3vccn1 triple mutant lines by transmission electron microscopy (TEM). No major changes  Fig. 11). Despite the broad distribution of data obtained from examination of 250-300 grana, the thylakoid stacks (grana) were slightly but significantly reduced in diameter in light-adapted as compared to dark-adapted plants within each genotype ( Supplementary Fig. 12). Within the same condition, only vccn1 had significantly increased granum diameter compared to wt, in accordance with previous observations 17 . Nevertheless, in contrast to the same report, the vccn1 grana were not curved in either light-or dark state ( Supplementary Fig. 11). In another previous study 14 , round-shaped chloroplasts often containing a large thylakoid-free stroma region were observed in dark-adapted clce to a higher extent than in wt. Such ultrastructural features could be also observed in the current study, but they were not typical of clce since they also appeared in the other genotypes especially in darkness. The reasons for not fully reproducing the previously-described phenotypes of vccn1 and clce are not clear at present. The different watering during plant growth (i.e., deionized water being used in this work and tap water with 1-5 mM Cl − in previous works) and different protocols for fixation and embedding of the samples between this and previous reports may impact the intra-thylakoidal ionic strength, www.nature.com/scientificreports www.nature.com/scientificreports/  Supplementary Fig. 9). Left and right panels in b and c: low-to-high and high-to-low light transitions, respectively. Data are the means ± SEM (n = 6-8 plants). Asterisks on green and red bars denote a statistically significant difference between wt and vccn1, and between wt and kea3 lines, respectively (one-way ANOVA test, P < 0.05). For size of proton motive force, see Supplementary Fig. 9. The experiment was repeated twice with similar results. (2019) 9:8639 | https://doi.org/10.1038/s41598-019-44972-z www.nature.com/scientificreports www.nature.com/scientificreports/ which was demonstrated to result in structural rearrangements in vccn1 17 . Taken together, the TEM data indicate wt-like thylakoid ultrastructure in all mutant lines, that could not explain the short-term alterations in NPQ and photosynthesis observed in this study.

Discussion
K + and Cl − fluxes across the thylakoid membrane have been predicted to function in membrane depolarization and regulation of photosynthesis [24][25][26] . In this study, we bring evidence that three molecular players in these fluxes, namely CLCe, KEA3 and VCCN1, act independently and by distinct mechanisms in the adjustment of photosynthesis to the light environment.

Distinct roles of VCCN1 and CLCe in the activation of photoprotection and electron transport.
VCCN1 was demonstrated to function as a voltage-dependent Cl − channel in electrophysiological experiments using planar lipid bilayers 17 , and was therefore proposed to mediate Cl − import into the thylakoid lumen. CLCe is believed to transport Cl − as well 14 , but to date, no electrophysiological evidence has been put forward for its substrate or mode of regulation. The mechanism of Cl − transport may be different since CLCe modulates photoprotection and electron transport in different manners and time scales from VCCN1 in the studied conditions. In dark-and growth light-adapted plants, CLCe is likely required for an optimal electron transport activity (Table 1). On transition from dark to low light, VCCN1 accelerates NPQ activation with a maximum at 0.5 min, time when CLCe further stimulates it with a maximum at 1 min of illumination (Fig. 3). In the remaining illumination time, when a steady state in NPQ is reached, CLCe but not VCCN1 stimulates electron transport through photosystems. On transition from dark to high light, VCCN1 alone accelerates NPQ activation with a maximum at 0.5 min until 3 min without largely affecting electron transport (Fig. 4).
VCCN1 and CLCe play different roles also in fluctuating light conditions, since VCCN1 acts in the activation of NPQ on transition to high light, whereas CLCe stimulates the electron transport on transition to low light (Fig. 5a). In support of distinct actions, knocking out both CLCe and VCCN1 genes resulted in a clce-like phenotype in electron transport and a vccn1-like phenotype in NPQ activation (Table 1, . The mechanism behind the faster activation of NPQ by VCCN1 involves dissipation of ∆Ψ and increase in ∆pH (Fig. 5b), supporting previous observations 16,17 . The positive effect of CLCe on electron transport takes place by a ∆pH-independent mechanism that could be related to Cl − homeostasis in the chloroplast (Fig. 2, Supplementary  Fig. 7 and ref. 14 ). Our results support the view of different modes of action for CLCe and VCCN1 in regulation of photosynthesis.

Role of KEA3 in the downregulation of photoprotection. While VCCN1 and CLCe had no influence
on NPQ relaxation, KEA3 accelerated this phase of photoprotection in conditions of decreasing light intensity (Fig. 5a). These are also the conditions at which KEA3 reduced PMF partitioning to ∆pH (Fig. 5b). Our observations support the model proposed by Armbruster and colleagues 13 where KEA3 is inactive in high light and is rapidly activated upon decreasing light intensity to export H + out of the lumen thus dissipating the pH gradient. KEA3 also favors the relaxation of the NPQ formed on transition from dark to low light, with a maximum at 1 min of illumination (Fig. 3). Thus, KEA3 overlaps in time with the action of CLCe which increases NPQ, resulting in additive effects in the clcekea3 double and clcekea3vccn1 triple knockouts. With respect to the electron transport, KEA3 stimulates it in the first 2 min of transition from dark to low light (Fig. 3), whereas CLCe also functions in steady-state growth light conditions ( Table 1). The likely mechanisms behind involve the dissipation of ∆pH by KEA3 and regulation of Cl − homeostasis by CLCe, as discussed above. These distinct mechanisms may explain the independent action of KEA3 and CLCe in the regulation of photoprotection and electron transport.
A model for the action of CLCe, KEA3 and VCCN1 in regulation of photosynthesis. On the basis of the kinetic pattern of parameters presented in Figs 3-5, we propose a model for the sequence of events in which CLCe, KEA3 and VCCN1 are involved in the regulation of photosynthesis (Fig. 6). In the first minute upon transition from dark to low light, VCCN1 pumps Cl − into the thylakoid lumen, resulting in a high pH gradient that activates NPQ 17 . A transient maximum is reached, also with the contribution of CLCe, which is closely followed by KEA3 relaxing the NPQ to a low steady state level while elevating the electron transport. The mechanism behind is that Η + in excess are exported by KEA3 in exchange for K + ions, resulting in a reduced pH gradient and relaxation of NPQ 12,13 . On transition from low to high light, VCCN1 elevates the pH gradient and re-establishes a high NPQ while adjusting the electron transport. On transition back to low light, KEA3 exports H + out of the thylakoid lumen and downregulates NPQ, while together with CLCe they reactivate the electron transport. The action of CLCe takes place by a mechanism involving Cl − homeostasis 14 . The presented model shows how three thylakoid channels/transporters work in an independent and sequential manner to adjust the distribution of ions on the two sides of the thylakoid membrane in ways that help plants to fine-tune photosynthesis according to the fluctuating light environment. Future validation of homologues genes found in other photosynthetic and non-photosynthetic organisms 10,27 will allow us to discover and understand widely conserved adaptive mechanisms to rapid changes in natural habitats.

Materials and Methods
plant material and growth conditions. The kea3-1 (SAIL_556_E12), clce-2 (SALK_010237) and vccn1-1 (SALK_103612) T-DNA insertion mutant lines were obtained from the Arabidopsis Biological Research Center (https://arabidopsis.org/abrc/) and were previously characterized 14,17,18 . Single mutant lines were crossed to generate double and triple mutant lines. For genotyping of the mutants, genomic DNA was isolated from 4-week-old plants and used for PCR. Gene-specific primers are listed in Supplementary Table 3 Leaf chlorophyll (Chl) content. Chl was extracted from leaf discs in 96% (v/v) ethanol at 65 °C for 10 min and its content was determined by spectrophotometry 28 . Measurements of Chl a fluorescence and P700 oxidation-reduction kinetics. Fast Chl fluorescence (OJIP) kinetics were recorded on 30 min dark-adapted plants by applying a saturating actinic red pulse of 3,600 µmol photons m −2 s −1 for 1 s using a Handy-PEA (Hansatech, UK) fluorometer. Where indicated, detached leaves were first incubated for 30 min in deionized water or in deionized water supplemented with 150 mM KCl under growth light conditions, then removed from the solution and dark adapted for 15 min before the OJIP kinetics were recorded 14 . The maximal quantum yield of PSII photochemistry (F v /F m ), the photosynthetic performance index based on equal absorption (PI ABS ), and the time to reach F m (t Fm ) were calculated using Hansatech PEA Plus v1.10 software according to Strasser and colleagues 29 . PI ABS consists of 3 partial performances, namely those related with the concentration of reaction centers, the maximal energy flux reaching PSII centers, and the electron transport at the onset of illumination 30 . Chl fluorescence was also measured on plants during steady-state illumination (4 h in growth light) using a MultispeQ instrument, and the PSII quantum yield (Y(II) was calculated according to Kuhlgert and colleagues 31 , (http://photosynq.org).
Slow kinetics of Chl fluorescence induction and P700 oxidation-reduction were simultaneously recorded with a pulse-amplitude modulated fluorometer DUAL-PAM 100 equipped with DUAL-DB and DUAL-E emitter-detector module (Walz, Effeltrich, Germany). The kinetics were first recorded on attached leaves of 30 min dark-adapted plants using actinic red light of 70 μmol photons m −2 s −1 (low light) or 660 μmol photons m −2 s −1 (high light) for 10 min. For experiments in fluctuating light, attached leaves of 30-min dark-adapted plants were first illuminated for 10 min with low light, then with 3 min of high light and then again 3 min of low light. NPQ and Y(II) were calculated based on changes in Chl fluorescence according to Genty and colleagues 32 , whereas Y(I) was calculated from absorbance changes at 830 nm (reflecting the redox state of P700) according to Klughammer and Schreiber 33 . The intensity of the saturating actinic red pulse applied to determine NPQ, Y(II) and Y(I) was 5,000 µmol photons m −2 s −1 and of 800 ms duration. NPQ, Y(II) and Y(I) differences were calculated as the difference of average value in wt minus average value in each mutant. Calculation of NPQ halftime (t 1/2 ) was performed in GraphPad. For this purpose, NPQ induction or relaxation data points were fitted with a single exponential decay and the decay constant (t1) was used to calculate the t 1/2 as follows: t 1/2 = t1 * ln(2), where ln (2) is natural logarithm of the number 2.
P700 oxidation-reduction kinetics were recorded based on absorbance changes at 830 nm using DUAL-PAM 100. The P700 signal calculated as the difference in transmittance at 875 nm and 830 nm is displayed as P700 Figure 6. Model for the regulation of photosynthesis by CLCe, KEA3 and VCCN1. The model illustrates the sequence of events during transition from dark to light and in fluctuating light. At the onset of illumination, photosynthetic electron transport generates a transmembrane charge polarity (voltage, −/+) which activates VCCN1. This allows a transient increase in NPQ without affecting electron transport. The function of CLCe resembles that of VCCN1 and partially overlaps in time with KEA3 action, which reduces NPQ to a low steady state while elevating the electron transport. On transition from low to high light, a new voltage is generated which reactivates VCCN1 resulting in increased NPQ. After transition from high to low light, KEA3 reduces NPQ while up regulating the electron transport in a process that also involves CLCe. VCCN1 and KEA3 modulate NPQ by changing the proton motive force partitioning to ΔpH (not illustrated in this model). The model is adapted from Fig. 4c  www.nature.com/scientificreports www.nature.com/scientificreports/ ΔI/I * 10 3 . Before the measurements, the plants were adapted to growth light for at least 1 h. The P700 signal was recorded during exposure of attached leaves to actinic red light of 120 µmol photons m −2 (AL, 5 min) and during exposure to far-red of 128 µmol photons m −2 (FR, 730 nm, 30 s) after 4 min in darkness. To monitor the P700 + reduction kinetics, one saturating actinic red pulse (20,000 µmol photons m −2 , 200 ms) was applied before AL was switched off and another one at the end of darkness and FR illumination. Where indicated, plants adapted to growth light were also used to record light response curves of Y(I) with 1 min exposure at increasing light intensity via P700 + -absorbance changes at 830 nm 33 .

Electrochromic band shift (ECS) measurements. ECS measurements were performed using a Walz
Dual PAM-100 equipped with a P515/535 module. First the plants were dark adapted for 30 min and then exposed to fluctuating light as described above. PMF size and partitioning into ∆pH and ∆Ψ were determined from a 60-s dark-interval relaxation kinetics of the ECS signal as described 5 at the end of the transition at the given light intensity. PMF size was calculated as the difference between the ECS signal in light and the minimum value of the ECS signal after the light was turned off (Supplementary Fig. 9). Calculation of ∆pH and ∆Ψ was performed using the steady-state time point of the ECS signal in darkness. Before each ECS measurement, a saturating 50-μs actinic red flash of 200,000 μmol photons m −2 s −1 was applied to determine the ECS ST ; subsequently, the ECS ST amplitude was used to normalize the ECS signal before the calculation of PMF size and partitioning values. transmission electron microscopy (teM). Leaves harvested from 7-week-old plants (age when the OJIP phenotype was observed in clce) that were 16-h dark adapted or 3 h after the onset of illumination were cut into small squares (2 × 2 mm) under dim green light and fixed in Karnovsky's solution (4% paraformaldehyde, 5% glutaraldehyde, 0.1 M sodium cacodylate buffer, pH 7.3) in vacuum for 4 h in darkness. The fixed leaves were washed three times with 0.1 M sodium cacodylate, pH 7.3. The leaves were post-fixed in 1% osmium tetroxide for 2 h. The samples were washed twice with 0.1 M sodium cacodylate buffer followed by water and thereafter dehydrated with a series of graded acetone. The leaves were infiltrated and embedded in Spurr's low viscosity resin. Leaf pieces of at least two different plants per treatment were analyzed independently. Ultrathin sections of 80 nm were made by a diamond knife using Leica (EM-UC7) or Reichert-Jung (Ultracut E) ultramicrotomes and mounted on formvar/carbon coated 200 mesh nickel grids. All the samples were contrasted by negative staining with readymade solutions of uranyless and lead citrate (Electron Microscopy Sciences, https://www.emsdiasum. com/) or by lead citrate. Electron micrographs were taken by Philips CM100 or JEOL JEM 1011 microscopes equipped with Morada high resolution digital camera. ImageJ software was used to measure granum diameter on the micrographs. Calculations were done on 250-300 randomly chosen grana originating from 20-25 chloroplasts taken randomly from 10-12 different mesophyll cells per treatment and genotype. The overall structure was investigated for >50 chloroplasts taken randomly from 15-20 mesophyll cells per treatment and genotype. statistical analysis. Statistics were performed in GraphPad software using the Tukey one-way ANOVA test.
For all data, P < 0.05 was considered significant.

Data Availability
The data that support the findings of this study are available from the corresponding author upon request.